Superior sodiophilicity and molecule crowding of crown ether boost the electrochemical performance of all-climate sodium-ion batteries

Significance

Sodium-ion batteries (SIBs) have received intense interest recently from academia and industry due to their abundant material resources. Despite remarkable progress at room temperature, SIBs still face challenges related to inferior charge transfer at low temperatures and severe side reactions at high temperatures. Herein, 15-Crown-5 (15-C-5) is screened as an electrolyte additive from a series of ether molecules theoretically and experimentally. The excellent sodiophilicity, high molecule rigidity, and bulky size of 15-C-5 allow it to reshape the solvation sheath and boost the electrochemical performance under all climate conditions. It is also widely adopted with various anode materials, demonstrating this screening protocol for advanced electrolytes.

Abstract

Sodium-ion batteries (SIBs) as one of the promising alternatives to lithium-ion batteries have achieved remarkable progress in the past. However, the all-climate performance is still very challenging for SIBs. Herein, 15-Crown-5 (15-C-5) is screened as an electrolyte additive from a number of ether molecules theoretically. The good sodiophilicity, high molecule rigidity, and bulky size enable it to reshape the solvation sheath and promote the anion engagement in the solvated structures by molecule crowding. This change also enhances Na-ion transfer, inhibits side reactions, and leads to a thin and robust solid–electrolyte interphase. Furthermore, the electrochemical stability and operating temperature windows of the electrolyte are extended. These profits improve the electrochemical performance of SIBs in all climates, much better than the case without 15-C-5. This improvement is also adopted to μ-Sn, μ-Bi, hard carbon, and MoS₂. This work opens a door to prioritize the potential molecules in theory for advanced electrolytes.

Sodium-ion batteries (SIBs) become appealing in the recent decade due to abundant material resources, similar operation fundaments to lithium-ion batteries, and unique electrolyte properties (1–5). As one of the most promising energy storage devices, SIBs need to maintain stable performance under various conditions, e.g., at high/low temperatures (6–10). However, inferior charge-transfer kinetics and increased side reactions in these cases significantly lower their performance (11, 12). Because these issues are closely related to electrolytes, the rational design of electrolytes is vital and attractive.

Using the low-temperature performance of SIBs as an example, the limited reports are based on weak/nonsolvation organics as the electrolyte solvent (WSE or NSE) (13–18). Wang et al. explored tetrahydrofuran (THF) as the electrolyte solvent to enhance the Na-storage performance of hard carbon (HC) at low temperatures (19). The weak solvation between THF and Na⁺ benefited the incorporation of anions in the solvation structures and the formation of inorganic-rich solid–electrolyteinterphase (SEI) on the anode. Meanwhile, it lowered the desolvation energy of Na⁺ and enhanced the reaction kinetics, especially at low temperatures. So, HC could deliver a specific capacity of 181 mAh g⁻¹ after 1,000 cycles at −20 °C. Lu and You reported diethylene glycol dimethyl ether (DGM) and THF as a mixed solvent to increase the electrolyte entropy and adapt to low temperatures (20). Using this electrolyte, HC delivered a reversible capacity of 250 mAh g⁻¹ after 500 cycles at −40 °C. However, using WSE or NSE would reduce the salt solubility, affecting the charge transfer properties of the electrolyte. Moreover, THF has a low boiling point (66 °C) and a low flash point (−14 °C) (21). The heavy use of THF (20 to 80 vol%) in the electrolyte abates its application potential, particularly at high temperatures. Hence, these works did not refer to the high-temperature performance. More importantly, electrolytes, especially ether-based electrolytes, show inferior electrochemical performance at high voltages. Therefore, they cannot couple the high-voltage cathode, which is essential for high-energy-density SIBs. Thus, it is highly urgent to expand the electrochemically stable window (ESW) of ether-based electrolytes. Up to date, the related reports in SIBs are rare, particularly for ether-based electrolytes.

Here, a variety of ether molecules are screened based on electrochemical stability, anion-engaged solvation, and basic physical properties in theory. As a result, 15-Crown-5 (15-C-5) stands out as an electrolyte additive to address the issues in the ether-based electrolyte, using NaPF₆ in DGM as an example. 15-C-5 has a good sodiophilicity with Na⁺, a high molecule rigidity, and a bulky size. All the features enable it to replace DGM in the solvation sheath and promote the anion engagement in the solvated structures by molecule crowding. In this context, the electrolyte shows enhanced Na-ion transportation, reduced side reactions, and thin/robust SEI on the anode. Meanwhile, ESW and operating temperature windows are greatly extended. Hence, these profits greatly boost the all-climate electrochemical performance of SIBs (−60 °C~60 °C), even using commercial Sn microparticles (μ-Sn) as the anode materials. The similar improvement is also identified in other anode materials, e.g., μ-Bi, HC, and metal sulfides. These results demonstrate the wide application of this electrolyte. The strong interaction, high molecule rigidity, and bulky size of 15-C-5 are totally different from the reported WSE or NSE. More importantly, this work opens a door to prioritize the potential molecules for advanced electrolytes in theory.

Results

Screening Ether-Based Molecules via Electrochemical Stability and Solvated Structures.

An ideal electrolyte needs to have a high ionic conductivity (>10⁻³ S cm⁻¹), a good electrochemical stability (0 to 5 V vs. Na⁺/Na), a low viscosity (1 to 10 mPa s), and a wide operation temperature (−30 °C~50 °C) as well. These metrics well guide the rapid screening of the potential electrolytes. Using ether-based electrolytes of SIBs as an example, the potential molecules are computed and identified based on theoretical calculations. Fig. 1A shows the potential ether-based molecules discussed in this work. Let us focus on electrochemical stability first because it is one of the most challenging issues for ether-based electrolytes. Goodenough proposed that the ESW of the electrolytes is related to the electronic states of the solvent and solvated cations, i.e., lowest unoccupied molecular orbitals (LUMO) and the highest occupied molecular orbitals (HOMO) (22, 23). The energy levels of HOMO and LUMO significantly change in the solvated structures, compared to their counterparts of free solvents (7). As calculated in SI Appendix, Fig. S1, the ESW of electrolytes can be associated with the difference in the energy levels of the LUMO of ether-Na+ and the HOMO of ether-anion. In general, a high energy level of LUMO and a low energy level of HOMO indicate a wide ESW. As summarized in Fig. 1B, 15-C-5 shows the best overall performance in these candidates, indicating its superior stability against the applied potentials.

Fig. 1.

Fast screening ether-based molecules for the electrolyte of SIBs. (A) The molecule formula and the abbreviations of potential ether molecules. (B) The energy levels of LUMO of ether-Na⁺ and HOMO of ether-PF₆⁻. (C) Molecule size and electrostatic potential (ESP_min) minimum of the ether molecules. (D) Dielectric constant and viscosity of the ether molecules. (E) Boiling point and melting point of the ether molecules.

The second metric is the sodiophilicity of these ether molecules. Usually, this term, which can be roughly estimated by the ESP minimum (ESP_min) of ether molecules, heavily affects the solvated structures (24). In our case, it is 15-C-5 that has the lowest ESP_min, reflecting the strongest interaction between it and Na⁺. The strong interaction enables it to reshape the solvation structures in the ether-based electrolytes. Meanwhile, the bulky size and high molecule rigidity of 15-C-5 result in a big interspace between neighboring molecules, thereby enabling anions to infiltrate and engage in the solvated structures. The anion-engaged solvation structures, contact ion pairs (CIPs) and ion aggregates (AGGs), are highly preferable for the anodes because they facilitate the formation of inorganic-rich SEI films and promote electrochemical performance (25, 26). In contrast, other ether molecules have either small molecule sizes or high molecule flexibility, forming a relatively dense solvation sheath around cations. Therefore, they would lead to an organic-rich SEI on the anode and inferior electrochemical performance. In short, there is an alternative way to achieve the anion engagement in the solvated cations by screening molecules based on good sodiophilicity, bulky size, and high molecule rigidity. 15-C-5 meets all the needs abovementioned (Fig. 1C).

The third metric for a good electrolyte is its physical properties, e.g., viscosity (η) and dielectric constant (ε). Generally, a low η and a high ε facilitate ion transport. 15-C-5 is good in terms of ε, but poor in terms of η. As elucidated in Fig. 1D, η of 15-C-5 (21.7 mPa s) is much larger than those of other ether molecules (0.5 to 3.5 mPa s). This result greatly affects the solubility of sodium salts and the ionic conductivity of the electrolyte (SI Appendix, Fig. S2). Therefore, 15-C-5 is used as an electrolyte additive, instead of an electrolyte solvent. The final metric to be considered is the melting point (T_m) and boiling point (T_f) of ether molecules. The higher T_f and the lower T_m indicate the wide operation temperature, expanding the application fields of SIBs. Fig. 1E summarizes the T_m and T_f of different ether molecules. None of these molecules satisfies the criteria. Compared to the other, DGM realizes the balance between high T_f and low T_m. Meanwhile, DGM is suitable as the electrolyte solvent in view of its energy levels of LUMO and HOMO. Based on these discussions, the electrolyte using 15-C-5 as an electrolyte additive and DGM as an electrolyte solvent is examined for SIBs by experiments.

Regulating the Solvation Sheath of Na⁺ by 15-C-5.

A series of spectroscopic characterizations were conducted to clarify these changes in the solvation structures induced by 15-C-5. Here, 1 M NaPF₆ in DGM was used as the benchmark and marked as the base electrolyte (BE). SI Appendix, Fig. S3, shows the ¹H-NMR spectra of the electrolytes containing different concentrations of 15-C-5. As the concentration of 15-C-5 increases, the resonances of DGM gradually move to the positions of free molecules. This result indicates that 15-C-5 replaces the solvated DGM molecules and the free molecules of DGM steadily increase. Compared to the NMR spectra, Electrospray ionization mass spectra (ESI-MS) gives us direct evidence of the change in the solvation sheath. As depicted in Fig. 2A, the dominant species in BE is [Na(DGM)]⁺ (m/z = 157.1). While with 15-C-5 in BE, the dominant species becomes [Na(15-C-5)]⁺ (m/z = 243.1), demonstrating again that 15-C-5 substitutes DGM to stabilize Na⁺. In the Raman spectra (Fig. 2B), the vibration of PF₆^‾ shifts from 762.8 cm⁻¹ of NaPF₆ to 733.9 cm⁻¹ of BE due to the salt dissociation in BE (27, 28). After adding 15-C-5 into BE, the vibration gradually turns back, indicating that the interaction between Na⁺ and PF₆⁻ recovers to some extent. In other words, CIPs/AGGs in the electrolyte likely increase due to 15-C-5. This conclusion is also supported by other experiments and theoretical simulations. In the ¹⁹F-NMR spectra, the downfield chemical shifts are detected with the increasing content of 15-C-5 (Fig. 2C), reflecting the reduced electronic density around F due to the formation of CIPs/AGGs (29). The content of CIPs/AGGs in the electrolyte can be roughly estimated by the Fourier transform infrared spectra (FT-IR) spectra. As illustrated in Fig. 2D, the peaks at 834.7 and 865.4 cm⁻¹ could be assigned to free anions and CIPs/AGGs (30). With the addition of 15-C-5, the content of CIPs/AGGs in the solvation structures increases from 37 to 68%. In theory, the high content of CIPs/AGGs would benefit the formation of inorganic-rich SEI and improve the electrochemical performance (31).

Fig. 2.

15-C-5 reshapes the solvation structures of Na⁺ in the electrolyte. (A) ESI-MS data, (B) Raman, (C) ¹⁹F-NMR, and (D) FT-IR spectra of the electrolytes (BE+ X% 15-C-5, X = 0, 2, 5, and 8). (E and F) three dimensional (3D) snapshots and (G and H) radial distribution functions of the solvation structures in two electrolytes, BE vs. BE+15-C-5 (8%). (I) Solvation energies of [Na(DGM)₂]⁺ and [Na(15-C-5)]⁺. (J) Binding energies of [Na(15-C-5)(DGM)]⁺ and [Na(15-C-5)(PF₆)].

Then, molecular dynamics (MD) simulations were carried out to understand the changes in the solvation sheath of Na⁺, using a moderate concentration of 15-C-5 in BE (8%) as an example. As shown in Fig. 2E, Na⁺ ions in BE are highly solvated by DGM. Usually, one Na⁺ ion is surrounded by two DGM molecules on average. The radial distribution functions (RDFs) indicate that O_DGM locates at ~2.4 Å away from Na⁺ (Fig. 2G), while F_PF6⁻ distributes in the range of 6 to 9 Å. The results exclude the engagement of PF₆⁻ in the primary solvation sheath. In other words, it suggests that Na⁺ and PF₆⁻ are separated by DGM, resulting in a typical structure of solvent-separated ion pairs. The average number of O_DGM near Na⁺ is ~6.0, agreeing well with the solvation sheath in Fig. 2E. In contrast, Na⁺ ions are surrounded by 15-C-5 and PF₆⁻ after adding 15-C-5 (Fig. 2F), i.e., a typical structure of CIPs. It is well consistent with the results from the Raman, NMR, and FT-IR spectra. In this CIP, both O_15-C-5 and F_PF6⁻ situate at the same distance to Na⁺ (~2.4 Å, Fig. 2H). Such a small distance suggests the enhanced interaction between Na⁺ and PF₆⁻, compared to the case of BE. To confirm this conclusion, we calculated the binding energy of Na⁺-2DGM, the dominant species in BE. As shown in Fig. 2I, the binding energy between Na⁺ and 2DGM is −5.13 eV, less than that between Na⁺ and 15-C-5 (−5.52 eV). It indicates the preferable binding between Na⁺ and 15-C-5. However, due to the small quantity and giant size of 15-C-5, two or more 15-C-5 molecules around one Na⁺ are too crowded. But one 15-C-5 molecule cannot saturate the coordination of Na⁺. Hence, some coordination sites must be available for other species, DGM or PF₆⁻. Given binding energy, it is PF₆⁻ that wins the competition and coordinates with Na⁺ (Fig. 2J). Therefore, DGM does not appear in the solvation sheath, although it is the electrolyte solvent.

The Changes in the Physicochemical Properties of the Electrolyte.

First, the influence of this reorganized solvation structure on Na⁺ transport is investigated. Based on the aforesaid results, the net charge of the solvation structure in BE+15-C-5 is reduced ([Na(15-C-5)(PF₆)] vs. [Na(DGM)₂]⁺). So, the electrostatic interaction on the solvation structure is alleviated, thereby facilitating Na-ion transportation. To confirm this conclusion, the mean-square displacement (MSD) of Na⁺ in two electrolytes was calculated at −20 °C. The sluggish charge-transfer kinetics is thought of as the rate-determining step of low-temperature performance. As presented in Fig. 3A, the Na⁺ diffusion in BE+15-C-5 is much faster than that in BE. It is also supported by the ionic conductivity of Na⁺ (σ_Na+) deduced from apparent ionic conductivity (σ) and transference number (t_Na+) at low temperatures. As the content of 15-C-5 varies from 0 to 8%, σ increases from 3.73 in BE to 5.49 in BE+15-C-5 (Fig. 3B) and t_Na+ rises from 0.14 in BE to 0.55 in BE+15-C-5 (Fig. 3C and SI Appendix, Fig. S4). Based on these data, the ionic conductivity of Na⁺ (σ_Na+) steadily increases with the content of 15-C-5 in BE+15-C-5 (Fig. 3D). The enhanced Na⁺ transportation reduces the electrode polarization and suppresses the parasitic consequences caused by anions.

Fig. 3.

15-C-5 changes the physical properties of the electrolyte. (A) The MSD of Na⁺ in two electrolytes. (B) Overall ionic conductivity (σ), (C) transference number (t_Na+), and (D) Na⁺ conductivity (σ_Na+) of BE+X% 15-C-5 (X = 0, 2, 5, and 8). (E and F) 3D snapshots and (G and H) RDFs of BE vs. BE+ 8% 15-C-5. (I) Electrolyte viscosity and (J) freezing points of BE+X% 15-C-5 (X = 0, 2, 5, and 8). Without special annotations, the physical properties were obtained at −20 °C.

Second, the influences of this reorganized solvation structure on viscosity and freezing point are investigated. In BE, Na⁺ and PF₆⁻ are separated in the electrolyte. One PF₆⁻ interacts with multiple DGM molecules via F_PF6⁻···H_(DGM) (Fig. 3E). This interaction readily leads to the condensation of DGM molecules, retards the ion diffusion and accelerates the electrolyte freezing, especially at low temperatures. While PF₆⁻ in BE+15-C-5 interacts with DGM and 15-C-5 (Fig. 3F). The RDF of F_PF6⁻···H in BE+15-C-5 shows a similar pattern to that in BE (Fig. 3G). However, in BE+15-C-5, F_PF6^- prefers to interact with H_15-C-5 (Fig. 3H). Therefore, DGM in BE+15-C-5 moves relatively freely, which is good to lowering electrolyte viscosity and freezing point of the electrolyte. The calculations are verified by experimental results (Fig. 3 I and J and SI Appendix, Fig. S5). The electrolyte viscosity decreases from 28.5 mPa s to 10.5 mPa s (Fig. 3I) and the freezing point declines from −17.1 °C to −58.4 °C, as the content of 15-C-5 increases from 0 to 8%.

Third, the influence of this reorganized solvation structure on chemical properties is also investigated. Using the hydrolysis reaction of PF₆⁻ as an example, PF₆⁻ in the solvation sheath faces a higher energy barrier than free PF₆⁻ (Fig. 4A), indicating that 15-C-5 remarkably inhibits this side reaction. The defluorination of PF₆⁻ is another common reaction in the electrolyte. Its chemical activity could be evaluated by the dissociation energies of P-F in PF₆⁻. As displayed in Fig. 4B, PF₆⁻ in the solvation sheath shows a smaller dissociation energy than free PF₆⁻, therefore benefiting the defluorination reaction and promoting the formation of fluorides in SEI. Such a result benefits the formation of a durable electrode/electrolyte interface. These theoretical calculations imply that the encapsulation of PF₆⁻ into the solvation structure by 15-C-5 effectively restrains the side reactions and promotes the formation of fluoride-rich SEI. This conclusion could be supported by the energy levels of HOMO and LUMO of the species in the electrolytes. As shown in Fig. 4C, the LUMO energy level of [Na(DGM)₂]⁺ is lower than that of free PF₆⁻ ions, indicating that the electrolyte decomposition in BE at low voltages is dominated by the solvation structure, [Na(DGM)₂]⁺. Thus, the as-obtained SEI is NaF-deficient and organic-rich. In BE+15-C-5, the LUMO energy level of [Na(15-C-5)(PF₆)] is also lower than that of free PF₆⁻ but slightly higher than that of [Na(DGM)₂]⁺. It indicates the enhanced stability of BE+15-C-5 at low voltages over BE. As to HOMO orbitals, the energy levels of free PF₆⁻ ions and [Na(DGM)₂]⁺ in BE are close but higher than that of [Na(15-C-5)(PF₆)] in BE+15-C-5, implying the worse stability at high voltages. This conclusion is supported by Linear sweep voltammetry (LSV) curves and High-Resolution transmission electron microscope (HRTEM) images. Using low-temperature batteries as an example, a small but clear oxidization peak at 3.5 V exists in the LSV curves of BE (SI Appendix, Fig. S6) but disappears in BE+15-C-5. It indicates that 15-C-5 effectively improves electrolyte stability at high voltages. As to the low voltages (<0.1 V), both of them show an increasing current, reflecting notable electrochemical reduction. But the anodes in the two electrolytes manifest a significant difference in the components and thickness of SEI. HRTEM images indicate that the SEI thickness on μ-Sn in BE+15-C-5 (~11 nm) is much thinner than that in BE (28.8 nm) (SI Appendix, Fig. S7), agreeing well with the elevated energy level of LUMO orbitals in BE+15-C-5.

Fig. 4.

15-C-5 changes the chemical properties of the electrolyte. (A) Energy barrier of PF₆⁻ hydrolysis, (B) dissociation energy of PF₆⁻, and (C) HOMO and LUMO energy levels of free PF₆⁻ and solvated PF₆⁻ ions. (D and E) High-resolution C1s and (F and G) F1s spectra of the SEI derived from BE and BE+15-C-5. (H) Young's moduli distributions of the electrodes cycled in BE and BE+15-C-5. (I and J) SEM images of the electrodes after cycles in BE and BE+15-C-5. (K) EIS spectra of the electrodes cycled in BE and BE+15-C-5.

The X-ray photoelectron spectra (XPS) spectra coupled with Ar-ion sputtering were acquired on the anodes made of commercial Sn microparticles (μ-Sn, SI Appendix, Fig. S8) to disclose the difference in SEI components. Fig. 4D shows the C1s spectra of the anode cycled in BE. The anode surface has a quantity of oxygenous species that are usually regarded as the by-products of electrolyte decomposition. While the intensities of these species notably decrease on the anode cycled in BE+15-C-5 (Fig. 4E). This contrast indicates that 15-C-5 significantly inhibits electrolyte decomposition, consistent with the theoretical calculations discussed above. In the F1s spectra, the content of Na_xPF_yO_Z remarkably decreases, but that of NaF increases for the anode cycled in BE+15-C-5 (Fig. 4G and SI Appendix, Table S1), compared to the case of BE (Fig. 4F). These results confirm the changes in the chemical activity of PF₆⁻ discussed above. The NaF-rich and organics-deficient features make the SEI generated in BE+15-C-5 robust and conductive during cycles, thereby facilitating the charge transfer and improving the mechanical properties. To confirm this conclusion, the Atomic force microscope (AFM) technique was used to characterize the mechanical properties of the anodes before and after cycles. SI Appendix, Fig. S9, shows the force responses of the anodes cycled in two electrolytes. The material hardening and fracture behaviors are clearly observed in the response curve of the anode in BE, but absent in the case of BE+15-C-5. The results indicate the high resistance of the anode to stress in BE+15-C-5. The comparison in Young’s modulus leads to the same conclusion. As displayed in Fig. 4H, the average Young’s modulus of the anode cycled in BE+15-C-5 is 6.5 GPa. It is similar to the data before cycles (SI Appendix, Fig. S10), but much higher than 3.6 GPa of the anode cycled in BE. The superior mechanical properties of the anode in BE+15-C-5 originate from the unique components of the SEI and benefit the structure stability upon cycles. The SEM image shows that the anode is uniform and closely attached to the current collector before cycling (SI Appendix, Fig. S11). After 30 cycles in BE, cracks emerge on the anode surface and the anode detaches from the current collector (SI Appendix, Figs. S12 and S13 and Fig. 4I). This result also accounts for why there is severe surface oxidization on μ-Sn in the case of BE (SI Appendix, Fig. S14). While for the anode cycled in BE+15-C-5, the negative effects are effectively avoided (SI Appendix, Figs. S12–S14 and Fig. 4J). The anode keeps the dense surface and firmly adheres to the current collector. Besides the mechanical properties, the difference in SEI also changes the charge-transfer kinetics at the interface. Electrochemical impedance spectra (EIS) suggest that both the ohmic impedance (R_s) and charge-transfer impedance (R_ct) in the case of BE+15-C-5 are smaller than those in BE (Fig. 4K). The small R_s reflects the enhanced ion transportation in BE+15-C-5. The small R_ct represents the fast charge transfer across the electrolyte/electrode interface owing to thin and NaF-rich SEI in BE+15-C-5.

In short, the influence of 15-C-5 on the electrochemical properties of the electrolyte could be illustrated in Fig. 5. First of all, 15-C-5 has a high binding affinity with Na⁺. So, it replaces DGM in the solvation sheath. However, the large size and molecule rigidity of 15-C-5 limit its number around Na⁺ so that the coordination number is not saturated yet. In this case, it is PF₆⁻, instead of DGM, to accomplish the solvation sheath of Na⁺, leading to the formation of CIPs. The change in solvation structures lowers the freezing points and promotes the ion transfer in the electrolyte. Meanwhile, the solvation structures suppress the hydrolysis of PF₆⁻ and promote the dissociation of PF₆⁻. They also alter the energy levels of HOMO and LUMO orbitals, expanding the ESW. In this case, the solvation structures, CIPs, benefit the formation of a NaF-rich, organic-deficient, and thin SEI. All the profits rooted in the change of solvation structures would greatly improve the electrochemical performance in a wide temperature range.

Fig. 5.

The underlying mechanism for enhanced electrochemical performance by reshaping the solvation structure with 15-C-5.

15-C-5 as an Electrolyte Additive Improves the Electrochemical Performance of SIBs.

Commercial Sn microparticles (μ-Sn) were selected as a model to clarify the changes in electrochemical performance induced by 15-C-5 because alloy-type anodes are usually considered one of the most challenging electrode materials. BE+8% 15-C-5 was chosen as the electrolyte of interest because it has the smallest viscosity, highest Na⁺ conductivity, and lowest freezing point. At a high concentration (≥10%), 15-C-5 will precipitate from the electrolyte (SI Appendix, Fig. S15). First of all, the basic electrochemical properties of μ-Sn are investigated at different temperatures. Fig. 6A shows the voltage profiles of μ-Sn in BE+15-C-5 at 0.1 A g⁻¹ in a temperature range from 60 °C to −60 °C. These profiles maintain similar contours and electrode overpotentials. At –60 °C, μ-Sn still delivers a reversible capacity of 459.4 mAh g⁻¹, 64.4% of the data at room temperature. At 60 °C, the capacity increases to 838.6 mAh g⁻¹, close to the theoretical capacity of Sn. Then, the low-temperature performance was investigated in detail. SI Appendix, Fig. S16, shows the dQ/dV curves of μ-Sn in two electrolytes at −20 °C. Compared to the case of BE, the redox peaks of μ-Sn in BE+15-C-5 are greatly enhanced, suggesting the increased electrochemical activity in this electrolyte. SI Appendix, Fig. S17, displays the initial discharge/charge voltage profiles of μ-Sn in two electrolytes at −20 °C. The reversible capacity of μ-Sn in BE+15-C-5 is ~649.1 mAh g⁻¹ at 0.1 A g⁻¹, much larger than that in BE (~234.4 mAh g⁻¹). Meanwhile, the initial Coulombic Efficiency (iCE) increases from 50.3% in BE to 70.9% in BE+15-C-5. The results imply that the irreversible reactions are notably inhibited. Fig. 6B manifests the rate performance of μ-Sn in two electrolytes at −20 °C. Obviously, the capacity in BE+ 15-C-5 is better than that in BE at all the current densities. The superior performance is due to the improved charge-transfer kinetics in BE+15-C-5, as discussed above. SI Appendix, Fig. S18, presents the cycling performance of μ-Sn in two electrolytes at −20 °C. At 0.1 A g⁻¹, the reversible capacity of μ-Sn in BE is always less than 100 mAh g⁻¹. It eventually fails after 30 cycles. In contrast, the specific capacity of μ-Sn in BE+15-C-5 remains at 647.1 mAh g⁻¹ after 100 cycles with a capacity retention of ~94.7%. The long-term cycling of μ-Sn in BE+15-C-5 was further examined at −20 °C. After 1,000 cycles at 0.5 A g⁻¹, μ-Sn exhibits a capacity of 505.3 mAh g⁻¹ (Fig. 6C), which is the best low-temperature performance for alloy-type anode materials to date (Fig. 6D and SI Appendix, Table S2) (32–36). The voltage profiles are kept well upon cycles (SI Appendix, Fig. S19). The electrochemical performance of μ-Sn in two electrolytes was also compared at room temperature. As depicted in Fig. 6E, μ-Sn presents stable cycling in BE+15-C-5 up to 4,000 cycles at 2 A g⁻¹, much better than in BE. The results are on the top list of the reported data (SI Appendix, Table S3). The enhanced electrochemical performance of μ-Sn also presents at 60 °C (Fig. 6F), where it keeps a capacity of ~766.9 mAh g⁻¹ in BE+15-C-5 after 100 cycles. The capacity of μ-Sn in BE drops to 98.5 mAh g⁻¹ after 50 cycles. To verify the effectiveness of molecule screening aforesaid, the electrochemical performances of the electrolytes using 12-Crown-4, dimethoxymethane, or tetraglyme as the electrolyte additives are also examined. Their performances are worse than that using 15-C-5, well supporting the proposed screening protocol (SI Appendix, Fig. S20). To exclude the improvement from Na metal in half cells, the symmetric cells of Sn||Sn are assembled. The symmetric cells of Sn||Sn in BE+15-C-5 present an extended cycle life in comparison to that in BE, confirming the positive contribution of 15-C-5 on Sn (SI Appendix, Fig. S21). Actually, 15-C-5 also enhances the electrochemical performance of Na metals (SI Appendix, Fig. S22).

Fig. 6.

Improved electrochemical performance by 15-C-5. (A) The galvanostatic voltage profiles of μ-Sn in BE+15-C-5 at different temperatures in half cells. (B) Rate performance and (C) cycling performance of μ-Sn in BE and in BE+15-C-5 at −20 °C in half cells. (D) Comparison of our results with the reported low-temperature performance based on the alloy-type anodes. (E and F) Cycling performance of μ-Sn in BE and in BE+15-C-5 at room temperature or at 60 °C in half cells. (G and H) Cycling performance of Bi, HC in BE and in BE+15-C-5 in half cells. (I) Galvanostatic voltage profiles, (J) rate performance, and (K) cycling performance of Sn||NVP@C using BE or BE+15-C-5 as the electrolyte at −20 °C. (L) Comparison of the reported low-temperature performance of SIBs with ours. (M and N) Cycling performance of Sn||NVP@C using BE or BE+15-C-5 as the electrolyte at room temperature and at 50 °C.

The benefits by 15-C-5 in the electrolyte are also observed in other anode materials. Let us use the low-temperature performance as an example. Fig. 6G shows the cycling performance of commercial Bi microparticles (μ-Bi, SI Appendix, Fig. S23) in two electrolytes at −20 °C. The anodes were discharged/charged at 0.1 A g⁻¹ for the first three cycles and then at 0.5 A g⁻¹ for the rest cycles. In this context, μ-Bi only survives for 23 cycles in BE. While in BE+15-C-5, μ-Bi exhibits excellent cycling stability. There is almost no capacity decay even after 200 cycles. Similar results are also verified for the electrochemical performance at room temperature (SI Appendix, Fig. S24). We also adopted this electrolyte for commercial HC (SI Appendix, Fig. S25), the most promising anode material of SIBs. At −20 °C, HC in BE only delivers a capacity of 56.2 mAh g⁻¹ after 200 cycles at 0.1 A g⁻¹ (Fig. 6H), but that in BE+15-C-5 presents a capacity of 203.2 mAh g⁻¹ under the same conditions. MoS₂ is used as a model of conversion-type anode materials to identify the effect of this electrolyte on its electrochemical performance. Homemade MoS₂ (SI Appendix, Fig. S26) delivers a capacity of 123.3 mAh g⁻¹ in BE after 50 cycles at −20 °C (SI Appendix, Fig. S27). Whereas, MoS₂ in BE+15-C-5 presents a capacity of 579.8 mAh g⁻¹ under the same conditions. All the results confirm that the electrolyte additive has good compatibility and widespread applications for various anode materials.

Finally, the full cells using BE+15-C-5 as an electrolyte were assembled and tested at different conditions. The homemade Na₃V₂(PO₄)₃@C (NVP@C) particles (SI Appendix, Fig. S28) were the cathode materials of the full cells (37). Then, the capacity ratio of the anode/cathode was 1:1.2. Fig. 6I depicts the voltage profiles of the full cells using different electrolytes at −20 °C. The reversible capacity in BE is only 79.1 mAh g⁻¹ at 0.1 A g⁻¹, where only the mass of the anode materials is taken into account. The iCE is as low as 36.4%. In contrast, the reversible capacity of the full cells using BE+15-C-5 is 595.2 mAh g⁻¹, almost eight times that in BE. Meanwhile, the iCE is promoted to 61.5%. The discharge voltage plateau is at 3.14 V and the voltage hysteresis is only 0.19 V. The improvement is also evidenced in the self-discharging behavior. After being stored in the fully charged state for 3 d, the full cells using BE+15-C-5 as the electrolyte maintain a capacity retention of 85.9% (SI Appendix, Fig. S29). While those using BE remain only ~70.4% (SI Appendix, Fig. S29). This contrast confirms the reduced side reactions in BE+15-C-5 again. In the rate performance, the full cells using BE+15-C-5 as the electrolyte exhibit a capacity of 259.6 mAh g⁻¹ at 1 A g⁻¹ (Fig. 6J). Those using BE deliver a capacity of ~100 mAh g⁻¹ at the same current density. The results reflect the enhanced reaction kinetics in BE+15-C-5. Fig. 6K shows the cycling performance of the full cells at 0.5 A g⁻¹ at −20 °C. In the case of BE, the full cells rapidly decay and become almost inactive after ten cycles. Whereas the full cells using BE+15-C-5 as the electrolyte present a capacity of 382.3 mAh g⁻¹ after 500 cycles with a capacity retention of ~100%. This result is much better than the reported data (Fig. 6L) (34, 38–41). A similar improvement in electrochemical performance by 15-C-5 is achieved at room temperature. As illustrated in Fig. 6M, the full cells using BE+15-C-5 exhibit a specific capacity of 643.2 mAh g⁻¹ after 200 cycles at 1.0 A g⁻¹ and a capacity retention of 94.6%. In contrast, the full cells using BE deliver a specific capacity of 286.9 mAh g⁻¹ and a capacity retention of 43.2% under the same conditions. SI Appendix, Fig. S30, presents the cycling performance of the full cells using BE+15-C-5 under harsh conditions, i.e., small N/P ratio (N/P = 1), lean electrolyte (16.0 μL mAh⁻¹), and high loading (4.0 mg cm⁻²). After the first three cycles at 0.1 mA cm⁻², the full cells show an areal capacity of 2.2 mAh cm⁻² after 100 cycles at 0.6 mA cm⁻². The enhanced electrochemical performance is also confirmed at 50 °C (Fig. 6N). The full cells using BE+15-C-5 retain a capacity of ~442.9 mAh g⁻¹ after 100 cycles, much better than the case using BE as the electrolyte (~148.6 mAh g⁻¹). The results confirm that 15-C-5 effectively improves the electrochemical performance in all climates.

Discussion

In conclusion, crown ether, 15-C-5, is introduced into the basic electrolyte (NaPF₆ in DGM) as an additive. Both experiment results and theoretical calculations reveal that 15-C-5 replaces DGM in the solvation sheath and promotes the formation of CIPs. The change in solvation structures lowers the freezing points and promotes Na⁺ transfer in the electrolyte. Meanwhile, it also affects the energy levels of HOMO and LUMO orbitals, extending the electrochemical stable window. Furthermore, the change in solvation structures suppresses the hydrolysis of PF₆⁻ but promotes the dissociation of P-F. As a result, the resultant SEI is NaF-rich and organic-deficient, making it robust, thin, and conductive. All these profits rooted in the change of solvation structures, would greatly improve the electrochemical performance of anode materials in a wide temperature range. The similar improvements are also identified in other anode materials, such as HC, μ-Bi, and MoS₂. It confirms the good application ability of this electrolyte for SIBs. Finally, the full cells using NVP@C as the cathode material, μ-Sn as the anode material, and BE+15-C-5 as the electrolyte exhibit excellent performance in a wide temperature range, demonstrating the promising potential of this electrolyte. More importantly, it opens a door about how to prioritize the potential molecules for advanced electrolytes. It may open a door for the rational screening of electrolyte additives.

Materials and Methods

Electrode Fabrications and Electrochemical Performance.

μ-Sn powders (≥99.5%, 325 mesh, Alab Chem. Tech.), polyvinylidene fluoride (PVdF, DoDochem), and acetylene black (AB, DoDochem) with a weight ratio of μ-Sn: AB: PVdF at 7:2:1 were ball milled for 4 h at 350 rpm. The obtained slurry was spread onto a copper foil by a doctor‘s blade, and the coated foil was dried at 60 °C overnight. After that, this foil was punched into circular discs with a mass loading of μ-Sn about 1.2 to 1.4 mg cm⁻². μ-Bi powders (99.99%, 200 mesh, Macklin), HC (Tianjin Annuohe New Energy Technology Co., Ltd.), and homemade MoS₂ (42) were used to prepare the electrode by a similar procedure to that of μ-Sn. The mass loading of these active materials was 1.2 to 1.4 mg cm⁻².

These discs as the working electrodes were assembled with a glass-fiber membrane (Whatman GF/F, GE) as the separator, homemade Na metal as the counter and reference electrode together into 2025-type coin cells. The processing was finished in a glove box (Mikrouna, Super 1220/750/900) filled with Ar (H₂O < 1 ppm, O₂ < 1 ppm). The cycling and rate performances were tested on multichannel battery-testing systems (Land CT2001A, China) within 0.01 to 1.0 V (vs. Na⁺/Na). The low-temperature operation was measured in temperature-controlling incubators (Yiheng Scientific Instrument, China). EIS were recorded on an electrochemical workstation (Autolab PGSTAT 302N, Switzerland) in the frequency range of 100 KHz to 0.01 Hz. LSV curves were measured on an electrochemical workstation (CHI 760D, Chenhua Instruments, China) in the voltage of 0.01 to 5 V at a scanning rate of 0.1 mV s⁻¹ with a two-electrode system, in which a stainless steel sheet was used as the working electrode and Na sheet was used as the counter electrode. All current densities and specific capacities are reported based on the mass of μ-Sn only. The transference number of Na⁺ (t_Na⁺) was achieved by the potentiostatic polarization of the symmetrical cells (43).

For full cells, Na₃V₂(PO₄)₃@C (NVP@C) was prepared by the reported method. Then, the electrode was prepared by mixing NVP@C, AB, and PVdF in a weight ratio of 8:1:1. The slurry was spread onto an aluminum foil after ball milling for 2 h at 350 rpm, and the electrodes were dried at 100 °C overnight. The capacity ratio of anode and cathode materials was approximately 1:1.2. The full cells were cycled at a current density of 0.1 A g⁻¹ for the first three cycles and then kept at a current density of 0.5 A g⁻¹.

Material Characterization.

SEM images, Focused ion beam column scanning electron microscope (FIB-SEM) images, X-ray powder diffraction (XRD) patterns, XPS, AFM, FT-IR, Raman, and ¹H-NMR spectra were characterized. More material characterization details are provided in SI Appendix.

Theoretical Calculations.

The Gromacs program was used to perform MD calculations. The Density of functional theory (DFT) calculations were carried out with the Gaussian 09W software package. More calculation details are provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Raw data have been deposited in Figshare (https://figshare.com/s/64d7ef67d3512f3f9e8a) (44). All study data are included in the article and/or SI Appendix.